In May 2020, Google introduced the concept of Core Web Vitals three new metrics to be added to the way Google measures user experience. All three metrics are more or less about perceived page speed: how fast it loads, how soon it becomes interactive, and how stable it is when loading.
Starting with RankBrain in 2015 and culminating in BERT in 2019, Google has made a huge leap in using AI to interpret both queries and search results. In just one year since BERT was launched, it went from being used in 10% to being used in nearly 100% of all English language queries.

BERT has proven incredibly effective in identifying the exact intent of a query. It can resolve lexical ambiguity, learn new words, correct misspellings, find synonyms, and account for previously disregarded stop words. Here is just one example of how BERT has improved search results from semi-relevant to completely.